You are on page 1of 12

BMC Public Health BioMed Central

Research article Open Access


Gender, school and academic year differences among Spanish
university students at high-risk for developing an eating disorder:
An epidemiologic study
Ana R Sepulveda*1,2, Jose A Carrobles1 and Ana M Gandarillas2

Address: 1School of Psychology, Autonomous University of Madrid, Spain and 2Epidemiology Department, Public Health Institute. Region of
Madrid, Spain
Email: Ana R Sepulveda* - a.sepulveda@iop.kcl.ac.uk; Jose A Carrobles - joseantonio.carrobles@uam.es;
Ana M Gandarillas - ana.gandarillas@salud.madrid.org
* Corresponding author

Published: 28 March 2008 Received: 29 July 2007


Accepted: 28 March 2008
BMC Public Health 2008, 8:102 doi:10.1186/1471-2458-8-102
This article is available from: http://www.biomedcentral.com/1471-2458/8/102
© 2008 Sepulveda et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract
Background: The aim of this study was to assess the magnitude of the university population at
high-risk of developing an eating disorder and the prevalence of unhealthy eating attitudes and
behaviours amongst groups at risk; gender, school or academic year differences were also
explored.
Methods: A cross-sectional study based on self-report was used to screen university students at
high-risk for an eating disorder. The sample size was of 2551 university students enrolled in 13
schools between the ages of 18 and 26 years. The instruments included: a social-demographic
questionnaire, the Eating Disorders Inventory (EDI), the Body Shape Questionnaire (BSQ), the
Symptom Check List 90-R (SCL-90-R), and the Self-Esteem Scale (RSE). The sample design is a non-
proportional stratified sample by academic year and school. The prevalence rate was estimated
controlling academic year and school. Logistic regression analysis was used to investigate adjusted
associations between gender, school and academic year.
Results: Female students presented unhealthy weight-control behaviours as dieting, laxatives use
or self-induced vomiting to lose weight than males. A total of 6% of the females had a BMI of 17.5
or less or 2.5% had amenorrhea for 3 or more months. In contrast, a higher proportion of males
(11.6%) reported binge eating behaviour. The prevalence rate of students at high-risk for an eating
disorder was 14.9% (11.6–18) for males and 20.8% (18.7–22.8) for females, according to an overall
cut-off point on the EDI questionnaire. Prevalence rates presented statistically significant
differences by gender (p < 0.001) but not by school or academic year.
Conclusion: The prevalence of eating disorder risk in university students is high and is associated
with unhealthy weight-control practices, similar results have been found in previous studies using
cut-off points in questionnaires. These results may be taken into account to encourage early
detection and a greater awareness for seeking treatment in order to improve the diagnosis, among
students on university campuses.

Page 1 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

Background boundaries of clinical relevance are still difficult to define


Eating disorders are considered the third most prevalent with accuracy.
chronic health condition among adolescent females [1].
Therefore, the epidemiology of eating disorders has Recently, there have been a number of studies that have
focused mainly on adolescents. However, low prevalence explored gender differences in eating disorders, while pre-
of clinical cases among the general population constitutes viously they have mainly focused on females [33-35].
the main difficulty of such study. Over the past decade, a However, a number of reports suggest that eating disorder
refined methodology has been used to obtain a preva- psychopathology and psychiatric comorbidity in males
lence rate near to 4.5% in Western countries [2-5], includ- and females are more similar than they are different [34].
ing Spain [6-10], even though there may be a higher rate As Carlat and Camargo [36] indicated that males account
when considering all sub-clinical cases. for 10% to 15% of all BN cases, with 0.2% of all adoles-
cents meeting the full criteria of bulimia nervosa. Specifi-
Eating disorders (ED), disturbed eating or body image dis- cally, studies carried out among males in college
satisfaction are not well documented in Spanish univer- populations showed that the frequency of eating disor-
sity populations, despite the fact that several studies have dered behaviours is also higher: 26% vomiting behaviour,
reported that these concerns and illnesses are present in use of laxatives, 41% binge [29]; 8% dieting, 10% binge-
this population [11-13]. Spanish university campuses do ing and 2% laxatives [26]; 5% bingeing [11]; and 10%
not have the facility to document reliably the number of reported fasting, diet pills or use of laxatives or vomiting
cases accompanied by a major or minor mental disorder, to lose weight [16]. This apparent contradiction suggests
or the needs and concerns related to student mental the need of further research in this area.
health. Nevertheless, there are sufficient results to indicate
that university students may have a higher proportion of Additionally, the cut-off points on the questionnaires are
unhealthy eating behaviours and attitudes and may be established based on the female population, therefore
considered a high-risk group [14-17]. they have not been validated for male samples and there
exists no specific items to assess other unhealthy behav-
A review of the existing literature reveals that there are sev- iours that might be associated specifically with males [36].
eral studies of college populations that observe the at-risk For example, Geist, Heinmaa, Katzman and Stephens [37]
prevalence for eating disorders to be between 0.9–3% in found differences in the type of body image preoccupa-
males and 7.3–18% in females (however the overall aver- tions among men and women, where males are less con-
age is around 11%). The majority of these studies esti- cerned with exact weight or clothing size and are more
mated prevalence using specific self-report questionnaires concerned with achieving an idealised masculine shape.
in which a cut-off score was established, for example, the The scoring on the EDI revealed that males exhibited a sta-
Eating Attitudes Test, EAT-40 [18] or EAT-26 [19] or on tistically significant lower drive for thinness and body dis-
some scales from the Eating Disorder Inventory (EDI) satisfaction than did females. Similarly, Joiner, Katz and
[20] as Drive for Thinness (DT) or Body Dissatisfaction Heatherton [38] found that the females scored highest on
(BD) [21-23]. Furthermore, a cut-off point on the EDI the drive for thinness scale, while on the contrary, males
total score has also been used as a screening tool, using a showed the most perfectionism and interpersonal distrust
cut-off point of 50, 40 or 43 [7,11,24,25]. Other studies in a sample of adolescents with bulimic symptoms. The
do not establish cut-off scores in the questionnaires and validity and discrimination of the EAT and EDI's subscales
only describe eating disordered behaviours such as binge- with cut-off scores has not been successful in screening the
ing, vomiting or laxative misuse, and other symptoms of male sub-clinical population [39].
clinical relevance (e.g., restrictive dieting, weight and
body image concerns). As these students fail to meet the On the other hand, the college environment also pro-
diagnostic criteria [16,26-29], it is difficult to compare motes high stress and anxiety that may contribute to
results between studies, in spite of the fact that those who future patterns of eating problems, especially in students
do not meet the specific criteria for an eating disorder may with competitive and perfectionism personalities [31,40-
nonetheless experience a significant amount of distress 43]. Futch and colleagues [41] indicated that medical stu-
related to their eating-disordered behaviour [30]. In addi- dents scored significantly higher on the DT scale com-
tion, several studies have demonstrated the predictive pared to students pursuing other avenues of study, such as
value of using self-report measures of eating habits as far history or sociology. This suggests that the prevalence of
as the risk to develop an eating disorder is concerned students at risk for an ED may increase with the competi-
[14,16,31,32] and in spite of the limitations, it seems to tiveness of the school or program.
reveal reliable indicators of unhealthy weight control
practices and harmful attitudes among populations where This study was carried out to explore the prevalence of
Spanish university students at high-risk for developing an

Page 2 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

eating disorder. The Eating Disorder Inventory (EDI-2) contacted personally by telephone or email the lecturers
[44] was used as a screening instrument to detect the high- to explain the study. Permission was given by each lec-
risk population based on previous results in a Spanish turer to administer an anonymous and voluntary battery
population. We acknowledge that the EDI is not a diag- of questionnaires in his or her classroom. The students
nostic instrument, however, it allows researchers to esti- were not reimbursed for their time (45 minutes) complet-
mate the proportion of the population within the ing the battery. Students were given the option of receiv-
spectrum of disordered eating behaviours and psycho- ing the results of their assessment if they gave a telephone
pathological traits that are closer to the clinical popula- number and coded name.
tion than the normal population. Specifically, the aims of
our study were the following: (1) to examine disordered Measures
eating behaviour and psychopathological symptoms by All participants completed the following questionnaires:
gender, (2) to estimate the total prevalence of the univer-
sity population at high-risk for developing an eating dis- Demographic Questionnaire
order by gender, academic year and school (3) to evaluate This questionnaire collected demographic variables (age,
variables associated with the population at high-risk for weight and height self-reported, marital status, parental
an eating disorder. Specifically, it is hypothesized that stu- education, employment status, cohabitation, psychiatric
dents that reach the overall EDI cut-off point will be asso- history) and information on health habits (weight control
ciated with (i) increased unhealthy weight-control compensatory strategies as dieting, vomiting or use of lax-
practices, (ii) increased body image dissatisfaction, atives, regularity of the menstrual cycle, time invested in
increased psychopathology and lower levels of self- exercise and/or sport, alcohol and/or cigarette consump-
esteem. Additionally, (iii) we expect differences by gender tion). Body mass index (BMI = weight (kg)/height (m)2)
when you compare at high-risk prevalence rate, (iv) we was calculated based on self reported height and weight.
expect differences by academic year when you compare at According to DSM-IV [45], a body mass index less than or
high-risk prevalence rate and (v) we expect differences by equal to 17.5 kg/m2, is considered as a diagnostic criteria
schools when you compare at high-risk prevalence rate. for anorexia nervosa.

Methods Eating Behaviour and Attitudes


Subjects and sampling method The Eating Disorder Inventory (EDI-2) [46] consists of 91
The study was conducted with university students items rated on a six-point scale (from 1 (never) to 6
enrolled in the first and fourth academic year at the (always)) that are divided into 11 subscales. The first 64
Autonomous University of Madrid, between October and original items are grouped into 8 scales and additionally,
April in the 2000–01 academic year. Of the 21 schools on 27 new items were added to form 3 more scales. It was
the campus, 13 schools with the highest number of stu- designed for the assessment of attitudinal and behav-
dents enrolled were selected. A total of 10,150 students ioural dimensions relevant to anorexia and bulimia ner-
were targeted. Two schools had placements outside of the vosa. The most extreme response is recoded as a 3, the
university in the fourth academic year, thus, students from immediately before is recoded as a 2, and the next
third academic year were included. To achieve a represent- response is recoded as a 1. The other three responses
ative sample of the university campus by academic year recodes as a 0. Scale scores are the sum of all items for each
and school, the sample design was proportionally strati- subscale. This questionnaire has good internal consist-
fied according to academic year and school, assuming a ency between 0.84 and 0.92 for each scale. The version
95% confidence interval and 0.05 of sampling error. A employed for this study was the Spanish version by Gar-
total of 4,682 students was identified as the desired sam- ner [44] which has also good psychometric properties.
ple size. The prevalence rate was estimated controlling Higher scores indicate higher disordered eating attitudes
academic year and school. and behaviours.

Procedure The findings of the two-stage epidemiologic study in a


This study was carried out in several university schools of Spanish population sample by Gandarillas and colleagues
the Autonomous University of Madrid by the author and (2003) [47] was taken into account, as these authors con-
a team from the School of Psychology. The team consisted ducted a precise validation of EDI in a non-clinical adoles-
of 10 psychology students in their last academic year who cent population. In the first stage the questionnaire EDI
were previously trained as interviewers. Once authoriza- was administered to a representative sample of 1534
tion was given by the University Dean to carry out the female school students between 15 and 18 years old. In
study, an invitation was sent by letter to each chosen the second stage, all the students took part in a clinical
school. Permission was given by the Deans of the different interview. Alpha coefficients were 0.92 for the total scale
University Schools for the ED epidemiological study. We and oscillated between 0.63 and 0.88 by subscales. The

Page 3 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

best cut-off point on the EDI total score (adding 8 original Statistical analysis
scales) was examined using receiver operating characteris- The results were analysed using frequency distributions by
tic (ROC) curve analysis which calculated the probability gender. Student data were t-tested for continuous varia-
of correct classification or prediction between clinical and bles and chi-squared tests were used for each categorical
non-clinical population. For a cut-off score > = 40 variable by gender. Prevalence estimates of the high-risk
(obtained by 27.1% of the sample), the test presents a sen- population for ED and corresponding 95% confidence
sibility of 86.1% (69.7–94.8), specificity of 74.9% (72.1– intervals (CIs) were calculated according to the academic
77.5). For a cut-off score > = 50 (obtained by 17.4% of the year and school by gender, once the design was deter-
sample), the test shows a sensibility of 72.2% (54.6– mined, according to a cut-off point on the EDI total score.
85.2), specificity of 84.5% (82.1–86.6). To discriminate A logistic regression analysis was used to investigate
male students at risk, an overall cut-off point of 40 was adjusted associations between school, academic year, gen-
selected. der while providing the proportion of the sample at high-
risk for an eating disorder, odds ratios (OR) and confi-
Body Image dence intervals. Statistical significance for gender and the
The Body Shape Questionnaire (BSQ) [48] consists of a high-risk population differences for categorical variables
34 item scale with scores between 1 (never) to 6 (always) were calculated using the chi-squared test. A Pearson cor-
for each item. It measures personal body dissatisfaction, relation coefficient, with the two-tailed test of signifi-
fear of gaining weight and the desire to be thin. The Span- cance, was used to assess relationships between scales.
ish version adapted by Raich and colleagues [49] was used Mean scores in the BSQ, SCL-90-R and RSE scales were
which has an internal consistency of 0.97. A higher score divided into quartiles according to high-risk or low-risk
indicates more body dissatisfaction. group and was calculated by gender. The highest quartile
(Q3) indicated higher scoring in the other scales. All p val-
Self-Esteem ues were two-tailed and statistical significance was set at p
Rosenberg's Self-Esteem Scale (RSE) [50] was used to < 0.05. Data was analysed with the program Statistical
assess the level of self-esteem. The RSE consists of 10 state- Package for Social Sciences (SPSS 10.0).
ments regarding a person's general beliefs about him/her-
self. Each item is measured on a four-point scale–from Results
strongly agree (3) to strongly disagree (0). Five items are Response Rates
reverse scored–from strongly disagree (3) to strongly agree Of the 4,682 students that were calculated as the opti-
(0) so that in each case scores go from less to more self- mum sample size, 2,551 students participated in the epi-
esteem. The RSE has high reliability (Cronbach's alpha = demiology study; the response rate for student
0.93) [51]. The Spanish version of the scale used for this participation was 54.5%. The main reasons for of lack of
study has good internal consistency with a coefficient participation were absenteeism and the teacher's non-
alpha of 0.88 [52]. Lower scores indicate lower self- presence during the data collection. A total of 2,386 stu-
esteem. dents were included for the data analysis based on the
number of returned valid questionnaires, of which 31.4%
General Psychopathology (n = 743) were men and 67.9% (n = 1620) were women.
The mental health of the sample was assessed using the The percentage of invalid questionnaires was 5.4% (n =
Symptom Check List 90 Revised (SCL-90-R) [53], Spanish 140), including those that did not answer two or more
version by Gonzalez de Rivera and colleagues [54]. It con- questions, and 0.9% (n = 25) were not taken into account
sists of 90 questions that gauge nine symptomatic dimen- because they did not fit the age range.
sions of psychopathology. The Global Severity Index
(GSI) indicates the level of psychological distress for each The Medicine and Psychology Schools provided the high-
individual. The internal consistency is between 0.81 and est response rates (100%) for the first year. Biology and
0.90. Higher score means higher psychological distress. Psychology Schools provided the highest response rates
(79.8% and 63.7%, respectively) for the fourth year. The
Definition of population at high-risk for an ED lowest response rate was provided by the Information and
The prevalence rate was estimated using the EDI question- Technology (IT) School (32% for the first year and 34%
naire as a screening tool. For the current study, students for the fourth academic year). The age range was from 18
who scored 40 or higher on the EDI total score in the to 26 years old, of which 62.8% (n = 1,479) belonged to
screening were defined as a high-risk population of devel- the first academic year and 29.8% (n = 709) belonged to
oping an eating disorder. the fourth, except for the Medicine and Teaching Schools,
of which 7.4% belonged to the third academic year (n =
175) due to the fact that they had placements outside of
the school in the fourth academic year. This group of stu-

Page 4 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

dents was coded as being in the fourth academic year for sonal sensitivity, depression and anxiety. The GSI score
purposes of data analysis. was greater for females than for males (mean = 0.72 (0.5)
vs. 0.66 (0.5); t = 2.3, p < 0.023). There were no significant
Socio-Demographic Variables differences by gender in the RSE (mean = 32 (5); t = 0.7, p
The mean age of the first academic year was 19 (SD = 1.6) = 0.50).
and the mean age of the students in the fourth academic
year was 21.8 (SD = 1.6). Females were the majority in Based on self-reported height and weight, the mean BMI
each group. Of all the students, 85.6% lived with both for men was 22.9 (SD = 2.6) and 20.6 for women (SD =
parents; 86% of the parents were married, 9% were 2.4). The mean desired BMI for men was 22.8 (SD = 1.8)
divorced or separated and 4.6% were widowed. No gender and 19.8 for women (SD = 1.5). Difference by gender was
differences were observed for the demographic variables. statistically significant (t = 20.0, p < 0.001). 1.3% (n = 9)
Gender differences were found for parental education, of the males and 6% (n = 83) of the females reporting a
where a higher number of male students reported having BMI of 17.5 or less. Dieting (9.3% vs. 18.6%; χ2 = 33.1, p
fathers with a university degree (43.3% vs. 28.9%; χ2 = < 0.05), laxatives use (2.9% vs. 5.5%; χ2 = 7.4, p < 0.05)
6.7, df = 2, p = 0.034). and vomit to reduce weight (2.6% vs. 4.6%; χ2 = 2.8, p <
0.05) was most prevalent for females. Amenorrhea was
Regarding the health habits of the students: 30% reported reported by 2.5% (n = 42) of the female sample where
they did exercise, with the majority of men indicating they only 7 of the females self-reported an ED. Only binge eat-
worked out at the gym (43.3%; 5 hours/week) and 50% of ing was higher in the male sample (11.6% vs. 7.2%, χ2 =
the women indicating they did aerobics (4 hours/week). 4.5, df = 1, p < 0.05). There were significant differences by
Thirty-four per cent of students engaged in sporting activ- gender for all of the behaviours described above.
ities; usually football for men (54%; 4.4 h/w) and swim-
ming for women (43%; 3.4 h/w). There was a significant Prevalence of University Students at High-Risk for ED by
gender difference in terms of the number of hours dedi- Gender, Academic Year and School
cated to doing exercise or sport (t = 4, p < 0.001). Thirty One percent of the total sample (n = 27) reported that
percent of the students smoked and women smoked more they were receiving or had received treatment for an eating
frequently than men (34% vs. 22%; χ2 = 33.6, df = 1, p < disorder. The prevalence of university students at high-risk
0.001). Binge drinking was more frequent among men for developing an eating disorder was 17.6% (19.6–16.1),
than among women (21.2% vs. 6.4%; χ2 = 119, df = 3, p 14.8% (18-11.6) for men and 20.8% (22.8-18.7) for
<0.001). women. There were significant differences by gender are
shown in Figure 1 (χ2 = 21.0, df = 1, p < 0.001).
Help-seeking at a mental health service: 10% of the males
and 12.5% of females reported having requested psycho- Regarding academic year and gender, 14.3% (18-10.5) of
logical or psychiatric services (χ2 = 3.8, df = 1, p = 0.05); the males and 18% (20.4-15.4) of the females in the first
32% of the sample did not specify the reason but those academic year were at greater risk for ED. In the fourth
students that answered reported depression (19%) and academic year, 15.4% (21.1-9.6) of the males and 22.8%
anxiety (14.5%) as the most common problems among (26.4-19.2) of the females were at greater-risk of develop-
men and depression (17.4%), anxiety (14.5%) and eating ing an ED. There were no significant differences by aca-
disorders (12.5%) as the most common problems among demic year (χ2 = 2.6, df = 1, p = 0.10).
women.
The prevalence of university students at high-risk for
Gender Differences in Psychopathological Symptoms and developing an eating disorder by school varied from
Eating Behaviour 11.2% to 28.2% and is presented in Figure 2. The lowest
Female students obtained higher drive for thinness (DT) prevalence rates were found in the IT School, at 11.2%
and body dissatisfaction (BD) mean subscale scores than (20.6-2), and the Economics School, at 12.8% (18.6-6.7).
males (t = 11.0 and t = 15.0, p < 0.001, respectively), while The highest prevalence rates were found in the Hispanic
the mean score in the bulimia (B) scale was greater for Literature School, at 28.2% (41.7-14.6), and the Law
males (t = 2.3, p = 0.02). Excessive concern about shape School, at 27.2% (34-20.3). However, there were no sig-
and/or weight was assessed by the BSQ, for which the nificant differences by school and gender (χ2 = 22.2, df =
females had significantly higher means compared with 12, p = 0.07). In each school, females had the highest
the males (mean = 73.1 (28) vs. 52.2 (18.5), respectively; prevalence rates.
t = 17.5, p < 0.001). There were significant mean differ-
ences by gender in the SCL-90-R scales (p < 0.001). Men A logistic regression was performed with an EDI cut-off
had higher scores for hostility, paranoia and psychosis point as the dependent variable and school, academic
and females had higher scores for somatisation, interper- year and gender as independent variables, and their vari-

Page 5 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

Figure
At high-risk
1 prevalence rate for an ED according by gender and academic year (95% C.I.)
At high-risk prevalence rate for an ED according by gender and academic year (95% C.I.).

ous interactions as predictors. Results showed that school ing an ED (OR = 0.99; 95% CI: 0.98 – 1.0; p = 0.12), while
(OR = 1.07; 95% CI: 0.94 – 1.21; p = 0.30) and academic gender was the only factor associated with the risk of
year (OR = 1.02; 95% CI: 0.89 – 1.15; p = 0.82) were not developing an ED.
associated with a significantly higher risk of developing an
ED, while gender was (OR = 1.16; 95% CI: 1.08 – 1.24; p
< 0.001). The interaction effect between the three varia-
bles did not emerge as a significant predictor of develop-

Figure
At high-risk
2 prevalence rate for an ED according by school (95% C.I.)
At high-risk prevalence rate for an ED according by school (95% C.I.).

Page 6 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

Gender Differences for Population at High-Risk for an ED the psychopathology index (r = 0.59, p < 0.001). In con-
Related to Psychopathology Symptoms trast, a statistically significant inverse association was
The percentages of unhealthy eating attitudes and behav- found between the overall score of self-esteem and the
iours amongst the high-risk population compared with overall score of EDI, BSQ and Global Severity Index (GSI)
the low-risk population by gender are shown in Table 1. (r = -0.53, p < 0.001, r = -0.39 and r = -0.56, p < 0.001,
All of these disordered eating patterns were more frequent respectively).
amongst the high-risk group, except for the body mass
index (BMI), where a greater number of women had a Figures 3 and 4 show the mean scores for the BSQ, SCL-
BMI of 17.5 or less (3.1% vs. 6.7%, χ2 = 5.4, df = 1, p = 90-R and RSE scales according to quartiles based on EDI
0.023). Both female groups presented statistically signifi- scores in males and females at high-risk or low-risk for an
cant differences in unhealthy weight-control practices ED (Q1: lowest scoring, Q3: highest scoring). Female and
compared with the male groups. Although, binge eating male students at high-risk were associated with higher lev-
behaviour was more frequent among male students. els of body dissatisfaction by BSQ and psychological dis-
tress (GSI) compared to students at low-risk for an ED.
Gender differences for high scores on EDI subscales were Gender differences were also found in the high-risk stu-
explored. Men scored significantly higher on the ineffec- dent group, where there was a trend toward higher scores
tiveness (t = 2.5, p = 0.02), perfectionism (t = 3.1, p = on the measure of psychological distress (GSI) in males at
0.01), interpersonal distrust (t = 4.6, p = 0.01), and matu- high-risk, while their female counterparts showed a simi-
rity fear subscales (t = 3.0, p = 0.01) compared with lar tendency, with higher scores on the BSQ scale.
females at high-risk for ED. However, there were no signif-
icant gender differences in the bulimia and interoceptive Discussion
awareness scores (t = 1.2, p = 0.3 and t = 1.1, p = 0.08). The first aim of this study was to determine the impact of
Females, by contrast, scored significantly higher in the gender-related differences in disordered eating behaviour
drive for thinness and body dissatisfaction when com- and psychopathological symptoms among university stu-
pared with high-risk males (t = -7.9 and t = -9.9, p < 0.05, dents. Results from female students indicate greater need
respectively). for concern related to eating behaviour, body dissatisfac-
tion and psychological distress compared to male stu-
A partial correlation was estimated for the overall score of dents. In addition, female students presented unhealthy
EDI and each scale after controlling for gender. A signifi- weight-control behaviours as dieting, laxatives use or self-
cant positive correlation coefficient was found for the induced vomiting to lose weight than males. A total of 6%
overall score of EDI and BSQ (r = 0.71, p < 0.001) and for of the females had a BMI of 17.5 or less and 2.5% had
amenorrhea for 3 or more months, meeting two of the

Table 1: Comparing disordered eating between low-risk and high-risk groups for an ED by gender

Unhealthy Eating Behaviours ED Low-Risk ED High-Risk


Males Females Males Females
N = 620 N = 1238 N = 99 N = 350 p

Behaviours
BMI < = 17.5 1.7% 9.9% 2.2% 3.2%a 0.001a
0.001*
I have gone on diets 7% 13% 22.4% 38%a 0.001a
0.001*
I have used laxatives to get thin 1.8% 2.7% 10.7% 14.6%a 0.001a
0.001*
Amenorrhea 3 months or more -- 1.5% -- 6.2%a 0.001a
0.001*
I frequently have binging episodes where I can not stop eating 7% 4% 26.3% 19.7% 0.001a
0.001*
I have vomited to get thinner 1.8% 1.4% 9.6% 16% 0.001a
0.000*
Eating Attitudes
Thinking about dieting 6% 21.2% 29.3% 70.8%a 0.0001a
0.001*
I am afraid to gain weight 7.9% 25.7% 43.4% 74%a 0.001a
0.001*
I have a great desire to be thin 7% 22.3% 27.7% 74%a 0.001a
0.001*

ap = between males and females of high-risk vs. low-risk.


*p = between males high-risk vs. low-risk/females high-risk vs. low-risk

Page 7 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

Figure 3 of the scales' scorings according to students at risk group IN MALES


Prevalence
Prevalence of the scales' scorings according to students at risk group IN MALES. Mean scores for scales according
to quartiles by EDI cut-off point.

Figure 4 of the scales' scorings according to student at risk group IN FEMALES


Prevalence
Prevalence of the scales' scorings according to student at risk group IN FEMALES. Mean scores for scales accord-
ing to quartiles by EDI cut-off point.

Page 8 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

diagnostic criteria for anorexia nervosa. In contrast, the encourage greater awareness of treatment availability
bulimia score was greater for males and a higher propor- [16,55].
tion of males (11.6%) reported binge eating behaviour.
Therefore, exploring gender differences was relevant in Statistically significant differences for gender were found
understanding the prevalence of unhealthy eating behav- among proportions of students at high risk for an ED, spe-
iours and attitudes amongst university students, accord- cifically 14.8% (18-11.6) for men and 20.8% (22.8-18.7)
ing to previous studies [11,16,17]. for women. This gender difference is apparently small
compared with other studies, where the at-risk prevalence
The second aim was to estimate the prevalence of a univer- in males is lower (e.g., 2.2% in Gandarillas et al., [56];
sity student population at high-risk of developing an eat- 3,3% in Ruiz et al., [10]). This may be due to the fact that
ing disorder using a large university representative sample cut-off points for the DT, B or BD scales and cut-off point
by gender, academic year and school. An overall cut-off on the EAT were used, and that males apparently fell into
point of 40 on the EDI questionnaire was used to deter- the high-risk category due to other factors. In our sample,
mine that 17.6% (19.6-16.1) of university students can be male students at high-risk also scored higher on the five
considered to be at risk of developing an eating disorder. scales related to psychopathological eating disorder traits,
Two main factors were taken into account, the decision to such as perfectionism or interpersonal distrust. This may
use an overall score as the screening point: a) the cut-off point to the development of general global vulnerability
points given on the DT or B scales are not sensitive enough for health more than scales related directly to weight and
to detect ED symptoms in male samples and b) the two- body shape attitudes and behaviours, which may be more
stage epidemiological study using the EDI questionnaire closely related to a trend of vulnerability in women. How-
among Spanish adolescents. Specifically, the previous val- ever, students of both genders exhibited lower self-esteem,
idation study of the EDI [47] was performed in a repre- which supports this pattern of vulnerability.
sentative female adolescent population where the sample
had previously participated in a clinical interview. Using a The third aim of this study was to estimate the unhealthy
cut-off score of 40, the authors observed that 27.1% of the eating behaviours that were associated with a population
sample was at high-risk of developing an ED which means at high risk for developing an ED. The population at high
that 31 clinical cases were detected from 42 cases. Two risk for an ED also exhibited more frequent weight- or
Spanish studies have used the global score of EDI (> = 50) shape-control attitudes and behaviours (dieting, use of
to screen the at-risk population for ED. The first study by laxatives, vomiting, etc.) than the population considered
Morande and colleagues [7] found that 12% of males and low risk. For example, several studies found that males do
31% of females were at risk in an adolescent population. not often engage in vomiting to control weight and/or
The second study by Lameiras and her team [11] reported their figure [26] but do engage in other at-risk behaviours,
that 6.7% of men and 6% of women were at risk of devel- such as steroid or hormone abuse [57]. Males in the high-
oping an ED in a university population. In addition, risk group reported more alcohol use and high-risk
Machado and colleagues [24] validated the Portuguese females reported smoking more cigarettes than the low-
version of the EDI and used a cut-off point of 43 to screen risk females. A higher percentage of males and females in
female college students [25]. The authors reported that the high-risk group indicated requesting psychological or
18.4% of the females reached this cut-off score, which is psychiatric services at some point. Neumark-Sztainer and
similar to the results of the current study. her team [58] found that adolescents with eating disor-
ders indicated increased smoking, binge drinking, illegal
In 1996, Becker and collaborators [55] carried out the first substance abuse, and suicide attempt rates. The data of
National Eating Disorders Program (NEDSP), which was this study confirm the observation by Fisher et al. [59]
conducted at more than 400 college campuses with 9,069 that unhealthy behaviours are grouped in vulnerable ado-
participants to detect disordered eating behaviors and lescents, with unhealthy eating attitudes frequently a part
provide secondary prevention. Although the overall prev- of this aggregation.
alence of clinical individuals was unknown, nearly three-
quarters of the sample received a recommendation for fur- Limitations of the Study
ther clinical evaluation. A follow-up random sample was Several factors must be taken into account when observ-
conducted 2 years later (N = 289), and it was found that ing these prevalence rates. Despite the sample size, sam-
nearly one half had made a first appointment with a coun- pling type and wide spectrum of assessed behaviours,
sellor for a clinical evaluation and 40% of those did seek there are several limitations. Firstly, the study was a cross-
treatment for ED. The prevalence of disordered eating was sectional study and thus, we are not able to infer causality.
higher than expected, indicating that ED symptomatology Another limitation is the use of self-report questionnaires
was widespread amongst university students. Thus, it is to collect data, as this method relies upon the honesty of
crucial to identify the high-risk subpopulation and the students. To mitigate this problem, questionnaires

Page 9 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

were conducted anonymously; however it is possible that disorders is high. A significant difference in prevalence
biased answers were collected, in which case the preva- rate has been found by gender but not by school or aca-
lence rate may have been underestimated. The response demic year. Unhealthy control-weight behaviours are
rate for student participation was considered satisfactory, associated to these vulnerable students, mostly amongst
even though it was lower than that of other studies carried females, with psychopathological symptoms presented as
out in a university sample (66% in Forman-Hoffman [16] part of this aggregation.
or 64% in Futch and colleagues [41]) but higher than the
37% response rate in the study conducted by Anstine and Specifically, female students presented more unhealthy
colleagues [21]. In this case, the result is due to the high weight-control behaviours to lose weight than males. A
absenteeism rate at public universities, where it is esti- total of 6% of the females met two of the diagnostic crite-
mated that approximately 40% of the students do not ria for anorexia nervosa. In contrast, the bulimia score was
attend any of the lectures, and 10% stop attending lectures greater for males and a higher proportion of males
after one month after starting the course. Attending lec- (11.6%) reported binge eating behaviour.
tures is voluntary and depends on the particular lectures
and lecturers. Students were not notified of the question- Assuming the prevalence of students at high-risk for an ED
naire session before the day of collection and the presence and from a preventive point of view, early detection of a
of the lecturer during data collection was also associated situation that poses risks and a better tailoring of treat-
with a higher rate of participation. ment to that situation is essential in order to improve the
diagnosis, especially among male populations.
Finally, another limitation of the study is related to the
chosen definition of 'population at high-risk for an eating Competing interests
disorder'. Bjomelv, Mykletun and Dahl [60] state that the The author(s) declare that they have no competing inter-
various definitions of an "eating problem" have a low ests.
degree of correlation and lead to variable prevalence. In
our study, students considered as at high-risk for an ED Authors' contributions
also presented higher prevalence of eating problems in ARS conceived and designed the study, oversaw all stages
females, although the prevalence among males was also of data collection and performed the statistical analysis
considerable. Nevertheless, clinical interviews were not and drafted the manuscript. JC coordinated all stages of
performed and consequently the accuracy of this preva- the study, gave feedback on design and reviewed the man-
lence cannot be assumed. uscript. AG revised the data analysis, interpretation of
results and reviewed the manuscript. All authors read and
The similarities of core eating disorder psychopathology approved the final manuscript.
and comorbid illness in male and female patients encour-
age the continued use of similar detection with both Acknowledgements
groups [61]. Nevertheless, precise items in questionnaires ARS was supported by a grant from Madrid Health Agency for carrying out
to assess unhealthy behaviours that might be specific this study (Orden 490/2000). Dr. Sepulveda is now supported post-doctor-
within male samples need to be developed [36]. ate Fullbright and Education Ministry Fellowship (EX2004/0481), Institute of
Psychiatry and Guy's Hospital, London, United Kingdom. We would like to
thank Professor Janet Treasure and Dr. Laura Southgate for their helpful
Further research is necessary to examine the possible rela- comments. We would also like to thank Professor Juan Botella for his val-
tionship between demanding environments and psycho- uable statistical advice during the study. We would like to acknowledge
somatic vulnerability, especially amongst university each Faculty's Dean and the teachers who gave me the permission and the
students. Health programmes have been recommended time to carry out this project. We are also grateful to the Psychology fifth
for educational services on university campuses [55]. In year students who helped me with the data collection and all the students
the light of this recommendation, our team has since car- who took part in the study.
ried out a healthy habits programme based on the
improvement of body image and self-esteem with a uni- References
1. Rosen DS: Eating disorders in children and young adolescents:
versity sample where positive results were obtained etiology, classification, clinical features, and treatment. Ado-
among female populations [56,62]. It is hoped that other lesc Med 2003, 14:49-59.
such efforts will soon be attempted in an effort to alleviate 2. Cotrufo P, Barretta V, Monteleone P, Maj M: Full-syndrome, par-
tial-syndrome and subclinical eating disorders: an epidemio-
potential risk factors and unhealthy behaviours and atti- logical study of female students in Southern Italy. Acta
tudes. Psychiatr Scand 1998, 98:112-115.
3. Favaro A, Ferrara S, Santonastaso P: The spectrum of eating dis-
orders in young women: A prevalence study in a general pop-
Conclusion ulation sample. Psychosomatic Medicine 2003, 65:701-708.
The results of this study suggest that the prevalence of uni- 4. Santonastaso P, Zanetti T, Sala A, Favaretto G, Vidotto G, Favaro A:
Prevalence of eating disorders in Italy: a survey on a sample
versity population at high-risk of developing an eating

Page 10 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

of 16-year-old female students. Psychother Psychosom 1996, amongst Iberian female college students. International Journal of
65:158-162. Clinical and Health Psychology 2004, 4:495-504.
5. Steinhausen HC, Winkler C, Meier M: Eating disorders in adoles- 26. Drewnowski A, Hopkins SA, Kessler RC: The prevalence of
cence in a Swiss epidemiological study. Int J Eat Disord 1997, bulimia nervosa in the US college student population. Am J
22:147-151. Public Health 1988, 78:1322-1325.
6. Gandarillas A, Zorrilla B, Sepulveda AR, Munoz E: Trastornos del Com- 27. Penas-Lledo E, Waller G: Bulimic psychopathology and impul-
portamiento Alimentario: prevalencia de casos clínicos en mujeres adoles- sive behaviors among nonclinical women. Int J Eat Disord 2001,
centes de la Comunidad de Madrid Madrid, Documentos Técnicos de 29:71-75.
Salud Pública; Instituto de Salud Publica; 2003. 28. Vanderheyden D, Fekken G, Boland F: Critical variables associ-
7. Morande G, Celada J, Casas JJ: Prevalence of eating disorders in ated with binging and bulimia in a university population: a
a Spanish school-age population. J Adolesc Health 1999, factor analytic study. Int J Eat Disord 1988, 7:321-329.
24:212-219. 29. Pyle R, Mitchell J, Eckert E: The incidence of bulimia in freshman
8. Perez-Gaspar M, Gual P, Irala-Estevez J, Martinez-Gonzalez MA, college student. International Journal of Eating Disorders 1983:75-85.
Lahortiga F, Cervera S: Prevalence of eating disorders in a rep- 30. Hay PJ, Fairburn CG, Doll HA: The classification of bulimic eat-
resentative sample of female adolescents from Navarra ing disorders: A community-based cluster analysis study. Psy-
(Spain). Med Clin (Barc ) 2000, 114:481-486. chological Medicine 1996, 26:801-812.
9. Rojo L, Livianos L, Conesa L, Garcia A, Dominguez A, Rodrigo G, San- 31. Fisher M, Schneider M, Pegler C, Napolitano B: Eating attitudes,
juan L, Vila M: Epidemiology and Risk Factors of Eating Disor- health-risk behaviors, self-esteem, and anxiety among ado-
ders: A Two-Stage Epidemiologic Study in a Spanish lescent females in a suburban high school. J Adolesc Health
Population Aged 12-18 Years. International Journal of Eating Disor- 1991, 12:377-384.
ders 2003, 34:281-291. 32. Patton GC, Selzer R, Coffey C, Carlin JB, Wolfe R: Onset of adoles-
10. Ruíz PM, Alonso JP, Velilla JM, Lobo A, Martín,A., Paumard C, Calvo cent eating disorders: population based cohort study over 3
AI: Estudio de prevalencia de trastornos de la conducta ali- years. BMJ 1999, 318:765-768.
mentaria en adolescentes de Zaragoza. Rev Psiquiatr Infanto- 33. Steiner H, Lock J: Anorexia nervosa and bulimia nervosa in
Juvenil 1998:148-162. children and adolescents: a review of the past 10 years. Jour-
11. Lameiras M, Calado M, Rodriguez Y, Fernandez M: Eating disorders nal Of The American Academy Of Child And Adolescent Psychiatry 1998,
among Spanish university students. Actas Esp Psiquiatr 2002, 37(4):352-359.
30:343-349. 34. Muise AM, Stein DG, Arbess G: Eating disorders in adolescent
12. Raich RM, Deus J, Munoz J, Perez O, Requena A: Estudio de las act- boys: a review of the adolescent and young adult literature.
itudes alimentarias en una muestra de adolescentes. Revista Journal of Adolescent Health 2003, 33:427-435.
de Psiquiatría de la Facultad de Medicina de Barcelona, 1991, 7:305-315. 35. Becker AE, Grinspoon SK, Klibanski A, Herzog DB: Eating disor-
13. Toro J, Castro J, Garcia M, Perez P, Cuesta L: Eating attitudes, ders. The New England Journal Of Medicine 1999, 340:1092-1098.
sociodemographic factors and body shape evaluation in ado- 36. Carlat DJ, Camargo CA: Review of bulimia nervosa in males. Am
lescence. Br J Med Psychol 1989, 62 ( Pt 1):61-70. J Psychiatry 1991, 148:831-843.
14. Field AE, Camargo CA Jr., Taylor CB, Berkey CS, Roberts SB, Colditz 37. Geist R, Heinmaa M, Katzman D, Stephens D: A comparison of
GA: Peer, parent, and media influences on the development male and female adolescents referred to an eating disorder
of weight concerns and frequent dieting among preadoles- program. Canadian Journal Of Psychiatry Revue Canadienne De Psychi-
cent and adolescent girls and boys. Pediatrics 2001, 107:54-60. atrie 1999, 44:374-378.
15. Shisslak CM, Renger R, Sharpe T, Crago M, McKnight KM, Gray N, 38. Joiner TE Jr., Katz J, Heatherton TF: Personality features differen-
Bryson S, Estes LS, Parnaby OG, Killen J, Taylor CB: Development tiate late adolescent females and males with chronic bulimic
and evaluation of the McKnight Risk Factor Survey for symptoms. The International Journal Of Eating Disorders 2000,
assessing potential risk and protective factors for disordered 27:191-197.
eating in preadolescent and adolescent girls. Int J Eat Disord 39. Rosen JC, Silberg NT, Gross J: Eating Attitudes Test and Eating
1999, 25:195-214. Disorders Inventory: norms for adolescent girls and boys.
16. Forman-Hoffman V: High prevalence of abnormal eating and Journal of Consulting & Clinical Psychology 56(2):305-8, 1988,
weight control practices among U.S. high-school students. 56(2):305-308.
Eat Behav 2004, 5:325-336. 40. Fredenberg JP, Berglund PT, Dieken HA: Incidence of eating dis-
17. Lock J, Reisel B, Steiner H: Associated health risks of adoles- orders among selected female university students. J Am Diet
cents with disordered eating: how different are they from Assoc 1996, 96:64-65.
their peers? Results from a high school survey. Child Psychiatry 41. Futch LS, Wingard DL, Felice ME: Eating pattern disturbances
Hum Dev 2001, 31:249-265. among women medical and graduate students. J Adolesc
18. Garner DM, Garfinkel PE: The Eating Attitudes Test: an index Health Care 1988, 9:378-383.
of the symptoms of anorexia nervosa. Psychol Med 1979, 42. Garner DM, Garfinkel PE: Socio-cultural factors in the develop-
9:273-279. ment of anorexia nervosa. Psychol Med 1980, 10:647-656.
19. Garner DM, Olmsted MP, Bohr Y, Garfinkel PE: The eating atti- 43. Rojo L, Conesa L, Bermudez O, Livianos L: Influence of Stress in
tudes test: psychometric features and clinical correlates. Psy- the Onset of Eating Disorders: Data From a Two-Stage Epi-
chol Med 1982, 12:871-878. demiologic Controlled Study. Psychosomatic Medicine 2006,
20. Garner DM, Olmsted MP, Polivy J: Development and validation 68:628-635.
of a multidimensional eating disorder inventory for anorexia 44. Garner DM: EDI-2, Inventario de Trastornos de la Conducta Alimentaria.
nervosa and bulimia. International Journal of Eating Disorders 1983, Madrid, TEA Ediciones; 1998.
2:15-34. 45. Association AP: Diagnostic and statistical manual of mental disorders 4th
21. Anstine D, Grinenko D: Rapid screening for disordered eating ed. Washington, DC, APA; 1994.
in college-aged females in the primary care setting. J Adolesc 46. Garner DM: Eating Disorder Inventory-2: Professional Manual. Odesa:
Health 2000, 26:338-342. Psychological Assessment Resources; 1991.
22. Moss RA, Jennings G, McFarland JH, Carter P: Binge eating, vom- 47. Gandarillas A, Zorrilla B, Munoz E, Sepulveda A.R., Galan I: Valida-
iting, and weight fear in a female high school population. J tion Of The Eating Disorder Inventory (EDI) In A Non-Clin-
Fam Pract 1984, 18:313-320. ical Population. Gaceta Sanitaria 2003, 17:Suppl 2:141.
23. Palmquist J, Berglund P, Dieken HA: Incidence of eating disorders 48. Cooper PJ, Taylor MJ, Cooper Z, Fairburn CG: The development
among selected female university students. J of the Am Diet and validation of the Body Shape Questionnaire. International
Assoc , 1995:64-66. Journal of Eating Disorders 1987, 6:485-494.
24. Machado PP, Goncalves S, Martins S, Soares I: The Portuguese 49. Raich RM, Mora M, Soler A, Avila C, Clos I, Zapater L: Revisión de
Version of the Eating Disorders Inventory: Evaluation of its la evaluación y tratamiento del trastorno de la imagen cor-
Psychometric Properties. European Eating Disorders Review 2001, poral y su adaptación en una muestra de estudiantes. Psicolo-
9:43-52. gemas 1994, 15:81-99.
25. Machado PP, Lameiras M, Goncalves S, Martins S, Calado M, Machado 50. Rosenberg M: Conceiving the self. New York: Basic Books New York:
BC, Rodriguez Y, Fernandez M: Eating related problems Basic Books edition. 1979.

Page 11 of 12
(page number not for citation purposes)
BMC Public Health 2008, 8:102 http://www.biomedcentral.com/1471-2458/8/102

51. Banos RM, Guillen V: Psychometric characteristics in normal


and social phobic samples for a Spanish version of the Rosen-
berg Self-Esteem Scale. Psychol Rep 2000, 87:269-274.
52. Echeburúa E: Manual práctico de evaluación y tratamiento de la fobia
social Barcelona, Martínez Roca; 1995.
53. Derogatis L: SCL-90-R. Administration, scoring and procedures Manual 1
for revised version of the SCL-90. Baltimore: John Hopkins University Press
Baltimore: John Hopkins University Press edition. 1977.
54. González de Rivera J, et al.: Manual del Cuestionario de 90 Síntomas
(SCL-90-R). Madrid: TEA Ediciones. Madrid: TEA Ediciones. edition.
2002.
55. Becker AE, Franko DL, Nussbaum K, Herzog DB: Secondary pre-
vention for eating disorders: the impact of education,
screening, and referral in a college-based screening pro-
gram. Int J Eat Disord 2004, 36:157-162.
56. Gandarillas A, Febrel C, Galan I, Leon C, Zorrilla B, Bueno R: Popu-
lation at risk for eating disorders in a Spanish region. Eat
Weight Disord 2004, 9:179-185.
57. Labre MP: Adolescent boys and the muscular male body ideal.
[References]. Journal of Adolescent Health 2002, 30:233-242.
58. Neumark-Sztainer D, Story M, Toporoff E, Himes JH, Resnick MD,
Blum RW: Covariations of eating behaviors with other health-
related behaviors among adolescents. J Adolesc Health 1997,
20:450-458.
59. Fisher M, Golden NH, Katzman DK, Kreipe RE, Rees J, Schebendach
J, Sigman G, Ammerman S, Hoberman HM: Eating disorders in
adolescents: a background paper. J Adolesc Health 1995,
16:420-437.
60. Bjomelv S, Mykletun A, Dahl AA: The influence of definitions on
the prevalence of eating problems in an adolescent popula-
tion. Eat Weight Disord 2002, 7:284-292.
61. Braun DL, Sunday SR, Huang A, Halmi KA: More males seek treat-
ment for eating disorders. The International Journal Of Eating Dis-
orders 1999, 25:415-424.
62. Sepulveda AR, Carrobles JA, Gandarillas A, Poveda J, Pastor V: Pre-
vention Program for Disturbed Eating and Body Dissatisfac-
tion in a Spanish University Population: A Pilot Study. Body
Image, 2007, 4(3):317-328.

Pre-publication history
The pre-publication history for this paper can be accessed
here:

http://www.biomedcentral.com/1471-2458/8/102/pre
pub

Publish with Bio Med Central and every


scientist can read your work free of charge
"BioMed Central will be the most significant development for
disseminating the results of biomedical researc h in our lifetime."
Sir Paul Nurse, Cancer Research UK

Your research papers will be:


available free of charge to the entire biomedical community
peer reviewed and published immediately upon acceptance
cited in PubMed and archived on PubMed Central
yours — you keep the copyright

Submit your manuscript here: BioMedcentral


http://www.biomedcentral.com/info/publishing_adv.asp

Page 12 of 12
(page number not for citation purposes)